Support for these was removed in 3721ff82ff as the feature is
inherently insecure. Checkpoints imply an extremely privileged
operation that is incompatible with the constraints of a
`podman run` command. However, we should still tell folks that
it no longer works with a clear error message. This is done
during Specgen processing, server side, which should be safer
than the previous client-side processing for these.
<MH: Fixed cherry-pick conflicts>
Signed-off-by: Matt Heon <matthew.heon@pm.me>
This reverts commit f4401567cd.
This commit has been identified as a serious security issue.
Running checkpoints via `podman run` introduces potentially
confusing behavior where all security for a container, even
options specified by the user at the command line, can be turned
off by the checkpoint. Checkpoints have special security
considerations and must be explicitly requested by the user to
ensure they are used safely. We recommend ensuring the lifecycle
of a checkpoint is fully monitored and integrity is ensured at
each stage to guarantee safety.
Signed-off-by: Matt Heon <matthew.heon@pm.me>
Now that the Deprecated markers are recognized, staticcheck (SA1019) flags the remaining uses of legacy fields we must keep for backwards compatibility (StaticIP/StaticMAC, CNINetworks, the deprecated k8s seccomp annotation constants). Add targeted nolint directives for those. Also fix a govet printf verb: the network backend is an integer, so use %d instead of %q.
Signed-off-by: Lokesh Mandvekar <lsm5@redhat.com>
gocritic's deprecatedComment check requires a blank comment line separating a symbol's description from its Deprecated: marker so tooling recognizes it. Add the missing separator lines.
Signed-off-by: Lokesh Mandvekar <lsm5@redhat.com>
golangci-lint 2.9.0 with go 1.26 makes the prealloc linter smarter and flags many slices that can be preallocated with a known capacity. Convert them to make([]T, 0, cap) and drop a now-unused nolint directive.
Signed-off-by: Lokesh Mandvekar <lsm5@redhat.com>
runWslCommand never returns err != nil because if there's an error, it calls Fail (which panics, so the code after it is unreachable).
Remove error returning and checking.
Inspired by the following linter warning:
> pkg/machine/e2e/config_windows_test.go:59:56: runWslCommand - result 1 (error) is always nil (unparam)
> func runWslCommand(cmdArgs []string) (*machineSession, error) {
> ^
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
(cherry picked from commit 94442ae44e)
Signed-off-by: Lokesh Mandvekar <lsm5@redhat.com>
The function logged the wrong error, logging ec which is the return value from ParseInt makes no sense since the error tells us the function failed so ec is invalid/undefined. Also ParseInt already includes the string input in the error so just drop it.
This was flagged by govet now that we use go 1.26 because %q was wrong for integers.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit 03ee16f5b4)
Signed-off-by: Lokesh Mandvekar <lsm5@redhat.com>
This is getting flagged with go.126 now. We need to use %d to log an integer.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit 526fbee8cc)
Signed-off-by: Lokesh Mandvekar <lsm5@redhat.com>
When parsing image envs we need to be strict about the format, only the
"key=value" format must be accepted. Just keys must be rejected as they
are not valid according to the image spec.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit 6c431b73db)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
With podman-remote we do not enter a our user namespace like we do with
local podman so we keep running with the real user id.
So if we then try to use chrootarchive as normal user it fails with:
creating mount namespace before pivot: operation not permitted
So simply revert back to the normal archive code.
Now the more interesting thing is we do have a test
"podman save to directory with oci format" but it never runs
rootless+remote in our CI system with our current matrix as we wanted to
reduce jobs.
So rethink the matrix and add one such job as this shows it is needed.
Fixes: 25aee24cbd ("use chrootarchive over plain archive package")
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit fd07b9c6ec)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
This is basically a partial backport of commit b14e833ef6
("machine: add test to check config mount").
It seems on the ubuntu github actions runners we have XDG_CONFIG_HOME
set so the tests do not actually use its own private dir but a shared
one for the configs which breaks the test assumptions.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Commit cf021c4ba9 switched the logic to also allow golangci-lint in
$PATH but forgot to update the call here.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit 13c6b5d786)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
The interfacer linter was removed a few years ago.
Remove the annotations as well.
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
(cherry picked from commit a7886d1fe8)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Using os.Is{Exist,NotExist,Permission} checks is not recommended in the
new code (see official documentation). While using it in the existing
code is OK, it may still result in a subtle errors later (for a specific
example of that, see [1]).
Replace those with errors.Is.
Generated by:
gofmt -r 'os.IsExist(a) -> errors.Is(a, os.ErrExist)' -w .
gofmt -r 'os.IsNotExist(a) -> errors.Is(a, os.ErrNotExist)' -w .
gofmt -r 'os.IsPermission(a) -> errors.Is(a, os.ErrPermission)' -w .
goimports -w .
git diff vendor test/tools/vendor | patch -p1 -R
[1]: https://github.com/opencontainers/runc/pull/5061
Signed-off-by: Kir Kolyshkin <kolyshkin@gmail.com>
(cherry picked from commit 7f7b35356f)
[Added extra replacements manually that were only in the 5.8 tree]
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
To avoid special chars from being interpreted.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit 571c842bd3)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
The server already does validate this so this is not strictly needed and
the client must trust the server no matter what.
But adding an extra check here does not hurt and may help prevent future
bugs.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit 12bec19426)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Just as additional hardening.
Note chrootarchive does not work on macos/windows, in that case it still
falls back to the regular pkg/archive.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit 25aee24cbd)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Under normal circumstances this is not a problem as the archive file
created podman container checkpoint will no create symlinks.
However if a user passes a custom archive they could contain symlinks
that point outside our root. To resolve them within the root use
securejoin.
Note this is not a security problem because the full archive must be
trusted by a user to begin with as it contain the full container config.
Fixes: #27977
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit abb5120624)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
The go std os package to will always make sure to use O_CLOEXEC, however
in cases where we directly call unix.Open() we need to pass that flag
explicitly.
I looked at this as there was a report of a leaked fd on the pasta list,
though I am not sure this will address it.
But anyway doing this should be rather safe and avoid leaks into other
processes.
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
(cherry picked from commit d20933df02)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
When creating a secret with driver=shell via the API, the file driver's
default DriverOpts (including path) were applied because DriverOpts was
empty. The shell driver rejects path as an unknown option, making it
impossible to create shell-driver secrets via the REST API or
podman-remote.
Only apply default DriverOpts from config when the requested driver
matches the configured default driver.
Signed-off-by: Joe Doss <joe@solidadmin.com>
(cherry picked from commit 667f55c2ce)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
When `podman-remote` tars the context, extra `podman-build-secret*` paths were
either dropped by `.dockerignore` (#25314) or archived as absolute paths so
`COPY . .` pulled host-shaped trees into the image (#28334).
Use relative names under the primary context for extra sources and do not apply
`.dockerignore` to those forced entries.
Fixes: https://github.com/containers/podman/issues/25314
Fixes: https://github.com/containers/podman/issues/28334
Signed-off-by: Jan Rodák <hony.com@seznam.cz>
(cherry picked from commit e18bb9ea8b)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
In the SystemCheck HTTP handler, when parsing the
unreferenced_layer_max_age query parameter fails, the error response is
sent but execution continues to `unreferencedLayerMaximumAge = &duration`
where `duration` is the zero value. This causes the system check to run
with a zero duration instead of returning the 400 error to the client.
Add the missing `return` after the error response.
Fixes#28350
Signed-off-by: crawfordxx <crawfordxx@users.noreply.github.com>
(cherry picked from commit 29d80a26af)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Setting Entrypoint= (empty value) in a quadlet .container file should
produce --entrypoint "" in the generated podman command, which clears
the image's default entrypoint. Previously this was silently ignored
because lookupAndAddString skips empty values.
Move the Entrypoint key out of the generic stringKeys map and handle
it separately so that an empty value is passed through.
Closes#28213
Signed-off-by: umut-polat <52835619+umut-polat@users.noreply.github.com>
(cherry picked from commit 926f562bd5)
Signed-off-by: Paul Holzinger <pholzing@redhat.com>
Previously, using --secret=id=foo,env=BAR in remote mode would fail because the client sent the env var name to the server, which tried to resolve it locally. This patch modifies the client to resolve the environment variable locally, write it to a temp file, and send it as a file-based secret.
Fixes#27494
Signed-off-by: ZuhairM7 <ZuhairM7>
Signed-off-by: ZuhairM7 <zuhairmerali@gmail.com>
- Update documentation: Differentiate `unless-stopped` from `always` - containers stopped by the user before a reboot will not restart.
- Add `should-start-on-boot` filter: Identify containers that require a restart after a system reboot.
- Update command documentation: Add `restart-policy` and `label!` filters to the documentation for container commands (rm, ps, start, stop, pause, unpause, restart).
- Add `restart-policy` and `shoud-start-on-boot` to completions.
- Update service: Update `podman-restart.service` to use the `needs-restart=true` filter.
- Preserve state: Preserve the `StoppedByUser` state across reboots.
- Update API: Add a `ShouldStartOnBoot()` method to the Container API.
- Update documentation: Add descriptions for the `should-start-on-boot` filter.
Fixes: https://issues.redhat.com/browse/RHEL-129405
Fixes: https://github.com/containers/podman/issues/20418
Signed-off-by: Jan Rodák <hony.com@seznam.cz>
<MH: Fixed cherry-pick conflicts>
Signed-off-by: Matt Heon <matthew.heon@pm.me>
- remove old CLI validation that only checked --pod flag
- add validation in namespaces.go to catch all paths (cli, quadlet, api)
- block userns mixing for all pods with infra, not just ipc/net
- update error message to be clearer
- fix test cleanup to use PodmanExitCleanly()
- use dynamic pod names in system tests to avoid conflicts
fixes#26848
Signed-off-by: 0xdvc <neilohene@gmail.com>
<MH: Fixed cherry-pick conflicts>
Signed-off-by: Matt Heon <matthew.heon@pm.me>
Fixes: #26588
For use cases like HPC, where `podman exec` is called in rapid succession, the standard exec process can become a bottleneck due to container locking and database I/O for session tracking.
This commit introduces a new `--no-session` flag to `podman exec`. When used, this flag invokes a new, lightweight backend implementation that:
- Skips container locking, reducing lock contention
- Bypasses the creation, tracking, and removal of exec sessions in the database
- Executes the command directly and retrieves the exit code without persisting session state
- Maintains consistency with regular exec for container lookup, TTY handling, and environment setup
- Shares implementation with health check execution to avoid code duplication
The implementation addresses all performance bottlenecks while preserving compatibility with existing exec functionality including --latest flag support and proper exit code handling.
Changes include:
- Add --no-session flag to cmd/podman/containers/exec.go
- Implement lightweight execution path in libpod/container_exec.go
- Ensure consistent container validation and environment setup
- Add comprehensive exit code testing including signal handling (exit 137)
- Optimize configuration to skip unnecessary exit command setup
Signed-off-by: Ryan McCann <ryan_mccann@student.uml.edu>
Signed-off-by: ryanmccann1024 <ryan_mccann@student.uml.edu>
Quadlets installed from `.quadlet` file now belongs to a single
application, anyone file removed from this application removes all the
other files as well.
Assited by: claude-4-sonnet
Signed-off-by: flouthoc <flouthoc.git@gmail.com>
Enable installing multiple quadlets from one file using '---' delimiters.
Each section requires '# FileName=<name>' comment for custom naming.
Single quadlet files remain unchanged for backward compatibility.
Assited by: claude-4-sonnet
Signed-off-by: flouthoc <flouthoc.git@gmail.com>
The remote client (podman --remote) was incorrectly throwing an error
when --detach-keys="" was specified for attach, run, or start commands.
According to documentation and the v1.7.0 release notes, specifying an
empty string should disable detaching, not cause an error.
Fixes: #27414
Signed-off-by: shiavm006 <shivammittal42006@gmail.com>
